Wang Z.,Connecticut Childrens Medical Center
International Journal of Biostatistics | Year: 2011
In disease classification, a traditional technique is the receiver operative characteristic (ROC) curve and the area under the curve (AUC). With high-dimensional data, the ROC techniques are needed to conduct classification and variable selection. The current ROC methods do not explicitly incorporate unequal misclassification costs or do not have a theoretical grounding for optimizing the AUC. Empirical studies in the literature have demonstrated that optimizing the hinge loss can maximize the AUC approximately. In theory, minimizing the hinge rank loss is equivalent to minimizing the AUC in the asymptotic limit. In this article, we propose a novel nonparametric method HingeBoost to optimize a weighted hinge loss incorporating misclassification costs. HingeBoost can be used to construct linear and nonlinear classifiers. The estimation and variable selection for the hinge loss are addressed by a new boosting algorithm. Furthermore, the proposed twin HingeBoost can select more sparse predictors. Some properties of HingeBoost are studied as well. To compare HingeBoost with existing classification methods, we present empirical study results using data from simulations and a prostate cancer study with mass spectrometry-based proteomics. © 2011 Berkeley Electronic Press. All rights reserved.
Hyams J.S.,Connecticut Childrens Medical Center
Expert Review of Clinical Immunology | Year: 2014
The treatment of Crohn's disease in children has undergone a revolution in the past decade following studies that have demonstrated the efficacy of anti-TNFα agents in producing durable clinical response/remission as well as reversal of growth delay in many patients. The positioning of biologic therapy continues to be debated. Should it be reserved for children failing conventional therapy including immunomodulators or should it be used as primary therapy shortly after diagnosis in children with more severe disease likely to suffer a more complicated disease course? Risk stratification will be crucial to any therapeutic decisions and emerging data hold promise that identification of those most likely to benefit will be available in the near future. © Informa UK, Ltd.
Rogers S.C.,Connecticut Childrens Medical Center
The journal of trauma and acute care surgery | Year: 2012
Motor vehicle occupant injury is a significant source of morbidity and mortality among children. Correctly used child safety seats (CSSs) substantially reduce injury morbidity and mortality. The objective of this study was to describe how parents learn to use and install CSS at newborn discharge. We prospectively enrolled maternal/newborn infant dyads at discharge from a large urban teaching hospital. Survey data included maternal demographics and parental knowledge on CSS installation. After survey completion, a certified child passenger safety technician observed and recorded CSS information, infant placement in CSS, and CSS placement in vehicle. Nine specific misuse categories were recorded. A total of 101 mothers were enrolled, with mean age 29.4 years (15-45 years); 52% were white, 18% were black, and 27% were Hispanic; 50% had college degree or higher; and 41% were privately insured. We observed 254 CSS errors (range, 0-7; mean, 2.5). There were 52% infant placement in CSS misuse errors (range 0-4; mean, 1.3), and 48% CSS placement in vehicle misuse errors (range, 0-4; mean, 1.2). The CSS placement misuse included 29% CSS not attached to vehicle. More frequent misuse occurred among non-white, non-college-educated mothers (p < 0.01).There was no difference in misuse related to how, when, and where mothers learned about CSS installation. Despite national, state, and hospital policies that require newborns to be transported in a CSS, we found a significant number of concerning CSS misuse in our study population. These results highlight the need for improved CSS education starting with the first ride home. Therapeutic study, level III.
Wang Z.,Connecticut Childrens Medical Center
Methods of Information in Medicine | Year: 2012
Background: Multi-class molecular cancer classification has great potential clinical implications. Such applications require statistical methods to accurately classify cancer types with a small subset of genes from thousands of genes in the data. Objectives: This paper presents a new functional gradient descent boosting algorithm that directly extends the HingeBoost algorithm from the binary case to the multi-class case without reducing the original problem to multiple binary problems. Methods: Minimizing a multi-class hinge loss with boosting technique, the proposed Hinge-Boost has good theoretical properties by implementing the Bayes decision rule and providing a unifying framework with either equalor unequal misclassification costs. Further - more, we propose Twin HingeBoost which has better feature selection behavior than Hinge-Boost by reducing the number of ineffective covariates. Simulated data, benchmark data and two cancer gene expression data sets are utilized to evaluate the performance of the proposed approach. Results: Simulations and the benchmark data showed that the multi-class HingeBoost generated accurate predictions when compared with the alternative methods, especially with highdimensional covariates. The multi-class Hinge-Boost also produced more accurate prediction or comparable prediction in two cancer classification problems using gene expression data. Conclusions: This work has shown that the HingeBoost provides a powerful tool for multi-classification problems. In many applications, the classification accuracy and feature selection behavior can be further improved when using Twin HingeBoost. © Schattauer 2012.
Bhandari A.,Connecticut Childrens Medical Center |
McGrath-Morrow S.,Johns Hopkins University
Seminars in Perinatology | Year: 2013
Bronchopulmonary dysplasia (BPD) is the commonest cause of chronic lung disease in infancy. The incidence of BPD has remained unchanged despite many advances in neonatal care. BPD starts in the neonatal period but its effects can persist long term. Premature infants with BPD have a greater incidence of hospitalization, and continue to have a greater respiratory morbidity and need for respiratory medications, compared to those without BPD. Lung function abnormalites, especially small airway abnormalities, often persist. Even in the absence of clinical symptoms, BPD survivors have persistent radiological abnormalities and presence of emphysema has been reported on chest computed tomography scans. Concern regarding their exercise tolerance remains. Long-term effects of BPD are still unknown, but given reports of a more rapid decline in lung function and their suspectibility to develop chronic obstructive pulmonary disease phenotype with aging, it is imperative that lung function of survivors of BPD be closely monitored. © 2013 Elsevier Inc.